Theoretical analysis of MOFs for pharmaceutical applications by using machine learning models to predict loading capacity and cell viability.

Metal organic frameworks (MOFs) have indicated great capacity and applications in drug delivery owing to their porous structures. Analysis of their drug loading capacity as well as cytotoxicity was carried out in this study via machine learning. The study employs a stacking regression approach to predict two critical outputs: Cell Viability (%) and Drug Loading Capacity (g/g) in MOFs. The proposed framework combines base models, including Multilayer Perceptron (MLP), Random Forest (RF), and Quantile Regression (QR), with a meta-model for enhanced accuracy and robustness. Principal Component Analysis (PCA) was applied to reduce dimensionality, and the Water Cycle Algorithm was used to optimize hyperparameters. Evaluation metrics, including R, Root Mean Squared Error (RMSE), and maximum error, indicated that the QR-MLP model outperformed the other models, achieving test R scores of 0.99917 for Drug Loading Capacity and 0.99111 for Cell Viability. These findings provide a new perspective of their chemical and biological uses since they show the efficiency of stacking ensemble approaches in handling challenging datasets and optimizing MOF-based drug delivery systems. Future work will focus on enhancing biocompatibility and optimizing drug release profiles to further improve the clinical applicability of MOF-based delivery systems.